MERIT-Infonomics Research Memorandum series. Pricing and Welfare Implications of Parallel Imports in the Pharmaceutical Industry

Transcription

1 MERIT-Infonomics Research Memorandum series Pricing and Welfare Imlications of Parallel Imorts in the Pharmaceutical Industry Catalina ordoy & Izabela Jelovac MERIT Maastricht Economic Research Institute on Innovation and Technology PO ox MD Maastricht The Netherlands T: F: htt://meritbbs.unimaas.nl International Institute of Infonomics c/o Maastricht University PO ox MD Maastricht The Netherlands T: F: htt://.infonomics.nl

2 Pricing and Welfare Imlications of Parallel Imorts in the Pharmaceutical Industry Catalina ordoy a and Izabela Jelovac b, a MERIT. Maastricht University. The Netherlands. b HOPE/EOZ. Maastricht University. The Netherlands. February 003 bstract: In this aer e investigate the imlications of ermitting arallel imorts of harmaceuticals roduced by a monooly, from one country to another. We use a model here countries differ in the atients level of co-ayment for buying harmaceuticals, and atients differ in the utility obtained from the consumtion of harmaceuticals. We sho that the effects of arallel imorts on total elfare are as follos: On the one hand, hen countries differ in their health system only, arallel imorts decrease total elfare; On the other hand, hen countries differ in the health needs of their atients only, arallel imorts enhance total elfare. Keyords: Parallel Imorts, Welfare. The authors ackoledge the helful comments of Pedro P. arros, Stefano Comino, Miguel Gouveia, Inés Macho-Stadler, Xavier Martínez Giralt, Pau Olivella, ndrés Perea, seminar articiants at Maastricht University, Universitat utònoma de arcelona, Università di Padova, articiants to the nd Health Economics Worksho (Lisbon, 00), and to the Fourth Euroean Conference on Health Economics (Paris, 00). Corresonding author. Deartment HOPE/EOZ FdGW. Maastricht University. P.O. ox MD Maastricht. The Netherlands. I.Jelovac@beoz.unimaas.nl

3 1. Introduction With this aer, e articiate to the ongoing debate over the benefits and drabacks from ermitting arallel imorts among countries. In articular, e investigate the ricing and elfare imlications of arallel trade of harmaceuticals beteen to countries. Parallel imorts are goods roduced genuinely under intellectual roerty right (IPR) rotection, laced into circulation in one market, and then imorted into a second market ithout the authorization of the IPR oner. They are identical to the legitimate roducts, excet that they may be ackaged differently and may not carry the original manufacturer s arranty (Maskus, 000). One imortant reason hy arallel imorts might arise, if they are ermitted, is to arbitrage aay international rice discrimination, hich is idely observed for harmaceutical roducts (see Maskus, 000). One exected effect of ermitting arallel imorts is a convergence in rices beteen countries. Ganslandt and Maskus (001) rovide evidence on rice convergence resulting from arallel trade of harmaceuticals in EU countries. The exected effect of arallel trade in terms of social elfare is not so clear-cut. Welfare is shon to either increase or decrease ith arallel imorts, deending on hether authors consider any of the folloing asects: different drug rices regulations across countries (Pecorino, 00); efforts of IPR oners to exert vertical rice control (Maskus and Chen, 00); the level of demand disersion across markets (Malueg and Schartz,

4 1994); and the need for manufacturers to recou their global research and develoment costs (Danzon, 1998). The main contribution of our aer is that it stresses the imortance of identifying the main determinants of international rice discrimination to understand the elfare effects associated ith arallel trade. We use a model that accounts for the differences beteen countries in terms of health insurance reimbursement olicies and in terms of drug needs reflected in the atients valuation for a drug. We neglect the effects associated ith different income levels across countries, even though this difference is likely to be an imortant determinant of international rice discrimination. When e consider differences in income only, the arallel imorts are exected to flo from lo-income countries toards high-income countries. Since arallel imorts generate rice convergence beteen countries, richer countries might benefit from arallel imorts hile oorer countries might be orse off (see Danzon, 1998). Hoever, international rice discrimination is likely to be caused not only by differences in income across countries, but also differences in other relevant characteristics of the demand. Otherise, ho could e anser the question raised by Maskus (001): Why might rices be higher in oor countries? 1 Characteristics of the demand that are esecially relevant for harmaceuticals rely both on insurance and on drug needs. oth can be secific to countries. On the one hand, the huge variations among national health systems can influence the ricing strategies of harmaceutical firms. In articular, the level of insurance reimbursement influences the ricing of drugs, since it directly affects the rice elasticity of the demand for drugs. If there 1 Maskus (001) reorts the finding that rices are elevated in such countries as South frica, Mexico and razil relative to those in Canada, Sain and Italy. 3

5 is no other regulation on drug rices as it is the case in Germany and in Denmark among other countries, the harmaceutical manufacturers ould charge higher rices in countries here insurance is more generous, taking advantage of a loer rice elasticity of demand. Pavcnik (00) rovides evidence on this relationshi beteen the atients co-ayment for buying drugs and drug rices in Germany. On the other hand, harmaceutical firms might also take advantage of differences in needs for a given drug among countries, charging higher rices here some endemic illness makes the need for the aroriate drug higher than in countries here this illness is not active, for examle. We tackle these issues using a model ith the folloing timing. In the first stage, a multinational monooly roducer sets the rice of a atented drug to be sold in to countries. If the rices are different beteen the to countries, arallel traders can buy, in the second stage of the game, drugs in the lo-rice country and re-sell them in the highrice country at a rice deending on hether the market for arallel imorts is monoolistic or cometitive. In the third stage of the game, the individuals in both countries choose to consume either one unit of the drug sulied by the monoolist, or one unit of the arallel imorted drug, or nothing, so as to maximize their utility. We first confirm a result already reorted in the literature: Parallel trade makes the rices converge beteen countries. s a reaction to the ossible entry of arallel traders in the market, the harmaceutical monooly roducer trades-off the benefits from rice discrimination ith the losses of facing cometition from arallel imorts in the high-rice country. Therefore, the monoolist increases the rice in the lo-rice country, and In the resence of additional rice regulations, hich is not the focus of our aer, this relationshi may not hold anymore. See Jelovac (00) for the effect of item-by-item negotiation on this relationshi. 4

6 decreases the rice in the high-rice country so as to deter some amount of arallel imorts. This does not mean that ermitting arallel trade results in global uniform ricing. Contrary to other aers (Malueg and Schartz, 1994, and Richardson, 00) in hich arallel imorts are assumed to imly de facto global uniform ricing, e obtain global uniform ricing only if consumers value the original drug and the arallel imorted drug equally. Hoever, as noted by Maskus (001), goods that are arallel imorted may not be erceived to be of the same quality beteen markets, even if the manufacturer laced them on the market originally, because of differences in ackaging or guarantees. This difference in ercetion leads in our model to the ersistence of some level of rice discrimination beteen countries, even hen arallel imorts are ermitted. Furthermore, e sho that the effect of arallel imorts on the total elfare is ambiguous. We identify to cases in hich the effect of arallel trade in terms of total elfare can be stated unambiguously. We sho that arallel trade increases the total elfare hen it takes lace beteen countries differing in their drug needs only. The rationale behind this ositive effect relies on the re-allocation of consumtion from individuals ith relatively loer needs in the exorting country, toards individuals ith relatively higher needs. The oosite re-allocation of consumtion is the result of arallel trade hen countries differ only in their health insurance reimbursement olicies. In that case, the total elfare decreases ith arallel trade. One secific feature of our model is orth mentioning in order to contrast our results ith an existing general result over the elfare effect of third-degree rice discrimination. In our model, a simle utility secification results in the same total quantity of drugs urchased, no matter hether arallel imorts are ermitted or not. n existing general result states that 5

7 rice discrimination reduces elfare if it does not increase total outut (see Tirole, 1988,. 138). Even though our aim is not to comare uniform ricing ith rice discrimination, our result dearts from this classical result: The elfare effect of loering rice discrimination by ermitting arallel imorts is not necessarily ositive, even though the total outut remains constant. It is the resence of differentiated co-ayment for buying harmaceuticals in our model, and their influence on ho the consumers surluses are comuted, that exlain this discreancy. 3 In the next section e describe the model. Section 3 discusses the equilibrium of the game. In Section 4 e analyze the elfare imlications of alloing arallel trade. Finally, e conclude in Section 5.. The Model We consider a multinational firm roducing a atented drug. We assume that the variable cost of roducing the drug is zero. The roducer acts as a monoolist given the atent on his roduct. He sells the drug in to countries, and, at rices and, resectively. If arallel imorts are ermitted, one or more holesalers can buy the drug in country i, i =,, at rice i, and re-sell it in the other country at rice, and at no cost excet the rice aid for the drug in the first country. 4 3 See Jelovac (003) for a more detailed analysis of the role of differentiated subsidies on the elfare effects of third-degree rice discrimination versus uniform ricing, in general. 4 For the role of transortation costs, see Ganslandt and Maskus (001). 6

8 Each country has a oulation hose size is normalized to one. For simlicity, individuals in both countries are assumed to have a utility additively searable in the consumtion of a numeraire comosite good and the consumtion x of the drug, ith x = { 0,1}. They have an income I at their disosal to buy the comosite good, and one or zero unit of the drug. In each country, individuals differ in their valuation of the drug,, hich is uniformly distributed on the suort [ i, i ] in country i, i =, : ~ U [ i, i ]. We assume, for simlicity, that i i = 1. To guarantee that the equilibrium solutions are interior, e also assume that 0, i =,. i Moreover, e assume that individuals refer to consume the drug sulied by the monoolist to the one sulied by the arallel imorter. Therefore, their valuation of the arallel imorted drug is eighted by < 1. This assumtion reflects the fact that, according to Maskus (001) among others, goods that are arallel imorted may not be erceived to be of the same quality beteen markets, even if they ere laced on the market originally by the manufacturer, because of differences in ackaging or guarantees. We assume that the exenses for drug consumtion of an individual, i x i, are artially reimbursed by some ublic health insurer in both countries, so that the individuals only ay a share i of it in country i. Therefore, the indirect utility function of an individual ith valuation in country i can be ritten as: { ; ; 0} U i = I + Max i i i, 7

9 if there are arallel imorts available in country i. Otherise, the utility function reduces to: U i { ; 0} = I + Max i i. We assume throughout the aer the folloing inequality: <, to account for the differences beteen the countries, and to guarantee that if arallel trade takes lace, it does so from country toards country. The timing of our game is assumed to be as follos. If arallel trade is ermitted, and assuming a riori that arallel trade takes lace from country toards country, then the monoolist sets the rices and in the first stage of the game so as to maximize his rofits: Π m = ( D + D) + D, here D i, i =,, stand for the demand for the drug directly sulied by the monoolist in country i, and D stands for the demand faced by the arallel imorter in the imorting country,. Then, in the second stage of the game, the arallel imorter sets the rice, as a Stackelberg folloer. If the arallel imorter is unique, he sets so as to maximize his rofit: Π = ( ) D. If there are many holesalers cometing ith each other in the arallel imorts market, then they set a rice equal to their marginal cost: =. In the third stage of the game, the individuals in both countries choose to consume either one unit of the drug sulied by the 8

10 monoolist, or one unit of the arallel imort if it is available, or nothing, so as to maximize their utility. If arallel trade is legally forbidden, then the second stage of the game reviously described vanishes, and D = 0. equilibrium. We solve the game by backards induction to derive the Nash sub-game erfect 3. The equilibrium of the game 3.1. enchmark: Parallel imorts are forbidden legally We first resent, as a benchmark case, the equilibrium of the game hen arallel imorts are legally forbidden. In the last stage of the game, individuals choose to consume either one unit of the good sulied directly by the monoolist in their country, or nothing. Given the utility: U i { ; 0} = I + Max i i, only the individuals in country i ith a valuation for the drug i i are going to buy one unit of the good. Therefore, the demand faced by the monooly in country i is: Di = i i i, i =,. Given these demands, the monoolist sets the rices and so as to maximize his rofit: Π m = D + D = ( ) + ( ). 9

11 The equilibrium rices that maximize this rofit ( and ) are resented in Table 1. The corresonding equilibrium demands (D and D ), rofit (Π m), consumers surluses (CS and CS ), and ublic exenses (PE and PE ) for aying a share 1 i, i =,, of the exenses associated ith drug consumtion, are also resented in Table 1. Table 1 Price Country = Country = Demand D = = D Cons. CS = I + surlus 8 CS = I + 8 Profit 1 Π = + m 4 Public exenses 1 PE = 1 PE = In both countries, the equilibrium monooly rices increase ith the ratio reresenting the maximum effective illingness to ay for the drug: 5 i, i =,. i 5 In this model, the maximum valuation for the drug is corrected by the co-ayment rate to obtain the maximum effective illingness to ay for the drug. 10

12 In articular, the monooly manufacturer can charge higher rices hen the co-ayment rate is loer, taking advantage of a loer rice elasticity of demand hen the co-ayment is loer. The equilibrium demands do not deend on the level of the atients co-ayment for buying the drug, since the rice faced by the individuals in both countries, i i, only deends on the maximum valuation for the drug in their country. Furthermore, the rice is loer in country, hich results from the assumtion that the maximum effective illingness to ay is loer in country. In this benchmark case, the monooly roducer discriminates as much as ossible the rices beteen the to countries. 3.. Parallel imorts are legally ermitted When arallel imorts are legally ermitted, the demands for both the arallel imort and the drug sulied by the roducer are realized in the third stage of the game. Parallel trade, if it takes lace, does so from country toards country by assumtion. Then, in country here the drug is not available as a arallel imort, the individuals ith a valuation buy one unit of the drug sulied by the monooly roducer in this country. Therefore, D =. In country here arallel imorts are available, only the individuals ith a valuation: ( ) ; 1, 11

13 1 maximize their utility buying one unit of the arallel imort: { } 0 ; Max. Therefore, = 0 ; 1 ) ( Max D, hich is equivalent to: =. ) (1 ) (, 0 if if D For arallel trade to be attractive to the individuals in country, e need a rice not only loer than, but loer than, to account for the fact that, ceteris aribus, individuals refer the drug sulied by the monoolist to the arallel imort. Individuals in country ith a valuation: 1 ( ), are better off buying one unit of the good sulied by the monoolist, if the arallel imort attracts some individuals in country, i.e. if. Otherise, individuals ith a valuation buy one unit of the good sulied by the monoolist in country. Therefore, the demand for the drug sulied directly by the monoolist in country is: =. 1 ) (, if if D

14 In the second stage of the game, the arallel imorter(s) can buy drugs in country, and decide uon the rice, anticiating the demands D, D and D reviously derived. If the arallel imorts market is a cometitive one, then the equilibrium arallel imort rice is: =. This rice attracts consumers in country only if =. That is, only if the difference in the rices charged by the monoolist in both countries is big enough. Otherise, i.e. if rices and are too similar, there ould be no room for the arallel imorters to enter the market in country. If there is only one monoolistic arallel imorter, then the equilibrium rice is the one that maximizes his rofit: 0 Π = ( ) D = ( ) ( ) (1 ) if if,. If the difference beteen and is high enough:, then the arallel imorter enters the market ith the folloing equilibrium rice : + =. Otherise, i.e. if >, there ould be no arallel imorts available in country. 13

15 14 In stage 1, the monooly roducer sets the rices and to maximize his rofit, anticiating the arallel imort rice and the demands D, D and D. The demand for the drug sulied by the monooly roducer in country is unaffected by the decision of the arallel imorter in stage. Therefore, the demand D that is anticiated in stage 1 is: D =. Given the sub-game erfect equilibrium in stage, the demand for arallel imorts in country that is anticiated in stage 1 is: < =, ) (1 ) (, 0 if if D henever the market for arallel imorts is cometitive or monoolistic. The demand D that is anticiated in stage 1 deends on the market for arallel imorts. If it is a cometitive market, then: < =. 1 ) (, if if D If it is a monoolistic market, then: < =. ) (1 ) ) ((, if if D Given these demands, the equilibrium rices and that maximize the roducer s rofit: m D D D + + = Π ) (,

16 are resented in Table. The corresonding equilibrium demands (D, D and D ), rofits (Π and Π ), consumers surluses (CS and CS ), and ublic exenses (PE and PE ), are also resented in Table. For the sake of clarity, e use the folloing notation: {, } m, c 0, ith: - = 0 = 0, if, and/ or if arallel imorts are legally forbidden. - m = ( ( ) + ), if >, arallel imorts are ermitted, and their market is monoolistic. - c = ( + ), if >, arallel imorts are ermitted, and their market is cometitive. The term allos us to resent the equilibrium solution in Table in an uniform ay, indeendently on the situation considered: either no market for arallel imorts, or monoolistic arallel imorts market, or cometitive arallel imorts market. Thus, in order to comare these three situations, it is enough to focus on, ith 0 < <. m c 15

17 Table Country Country Parallel imorter Price 1 = + 1 = = Demand D = D = D = Cons. surlus ( ) CS = + ( + ) CS = CS CS Profit Πm = Πm ( ) Π = 1 + Public exenses 1 PE = 1 ( ) PE = ( ) If, then alloing arallel imorts or not does not make any difference, since = 0 = 0. In that case, the market conditions in both countries are very similar. That can be seen adding our assumtion on the market asymmetry: <, to the condition characterizing the situation discussed no: <. With such a similarity beteen the market conditions of both countries, the room for the monooly roducer to rice discriminate is very limited, no matter hether arallel imorts 16

18 are tolerated or not. Therefore, no arallel imorter could take advantage of this rice difference to attract clients in country. If >, market conditions in both countries are different enough, so that arallel trade occurs if it is alloed. We can discuss the effects of alloing arallel imorts, indeendently on hether the arallel imorts market is cometitive or monoolistic. In both cases, > 0, hich can be comared to the benchmark situation here arallel imorts are forbidden and = 0. In Table, e see that alloing arallel imorts makes the rices in both countries converge : increases and decreases. This is an intuitive result, and it is exlained by the folloing trade-off faced by the monooly roducer. The latter ould like to enjoy the benefits associated ith the rice discrimination, and limit the cometition associated ith the arallel trade in country. The trade-off arises because the stronger the rice discrimination, the bigger the room for arallel imorts. The main difference beteen a cometitive arallel imorts market and a monoolistic one can be understood hen realizing that m < c. This imlies that the rice convergence due to arallel imorts is stronger hen the arallel imort market is cometitive. This haens because the afore-mentioned trade-off and its resulting rice effect are stronger hen the threat of cometition from arallel imorters is stronger, thus hen the arallel imorts market is cometitive. Consequently, all the remaining effects associated ith arallel imorts are stronger hen the arallel imort market is cometitive. 17

19 The rice set by the arallel imorter is naturally higher or equal than the rice aid in country, and it is loer than the rice of the cometing drug sulied by the roducer in country. Given the convergence in rice, e have that: < < <. Therefore, individuals in country enjoy loer rices hen arallel imorts are ermitted, hile individuals in country face a higher rice. nalyzing the demands in Table, e deict a re-allocation of the drug consumtion from country to country, the total outut remaining unchanged. This can be seen grahically in Figure 1: Figure 1 Country : Α Country : Clearly, the consumers in country are better off, enjoying more consumtion at loer rices. The oosite haens in country. This exlains hy CS is loer and CS is higher hen > 0 than hen = 0. 18

20 The monooly roducer rofit is reduced due to both the cometition from arallel imorts in country, and the loer rice discrimination. The rofit of the arallel imorter(s) is obviously at least as high as hen they do not oerate on the market. ermitted. Last, the ublic exenses are loer in both countries hen arallel imorts are 4. The elfare analysis We no analyze ho the changes induced by the arallel imorts affect the total elfare. We define the total elfare as the sum of consumers surluses net of the ublic exenses in both countries, and rofits of both the monooly roducer and the arallel imorter(s): TW = CS + CS PE PE + Π m + Π. We already kno that, on the one hand, arallel imorts, hen they take lace, cause a ositive effect on the total elfare through CS, PE, PE, and eventually Π (if the arallel imorter is a monooly; otherise, Π = 0). On the other hand, they have a negative effect on the total elfare through CS and Π m. In order to determine the circumstances under hich the ositive effect out-eights the negative one, it is useful to comare the total elfare hen arallel imorts are alloed, TW, ith the one characterizing the benchmark case, TW : TW = TW + ( ( 1+ ) ), 19

21 here: TW = CS + CS PE PE + Π. m Therefore, the necessary and sufficient condition for arallel imorts to increase total elfare is: > ( 1+ ). [NSC] Given our assumtions, and the condition for the arallel imorts to take lace ( > ), arallel trade can result either in an increase or in a decrease of the total elfare. The sum of the rofits net of the ublic exenses alays decreases as a result of arallel trade: j= m, Π j i=, j= m, j PE = Π PE ( 1+ ). i i=, i Therefore, the total elfare can increase ith arallel trade only hen the gain for the consumers in country is sufficiently large to comensate the loss for the consumers in country. necessary but not sufficient condition for TW > TW is thus: i=, CS i = i=, CS i + ( + (1 + ) ) > CS, i=, i hich is equivalent to: + ( 1+ ) > 0. [NC] oth conditions [NC] and [NSC] hold hen countries only differ in the distribution of valuations for the drug, reflected in i, i =,. This haens hen e consider countries ith a similar health system, but ith different valuations for the drug due to 0

22 differences in the endemic illnesses suffered by their oulations, for examle. In that case, the condition under hich arallel imorts take lace reduces to >. Therefore, in this situation, the increase in the consumers surlus in country more than comensates the decrease in the one of country. One exlanation for that relies on the re-allocation of the drug consumtion from country toards country. The arallel imorts ould make the individuals in country ith a valuation:, +, give u consuming the drug. While in country, individuals ith a valuation:,, start consuming the drug thanks to the arallel trade. Therefore, e have a re-allocation from individuals valuing the drug less toards individuals valuing the drug more, since: + <, henever { } m, c. This intuition can be seen grahically in Figure, hile Proosition 1 summarizes this case. 1

23 Figure Country : Α + Country : Proosition 1. Parallel imorts increase the total elfare hen they take lace beteen countries differing only in the distribution of the valuations for the drug among their oulation: = and < TW TW. nother interesting case considers to countries differing only in their health care system, reflected in the co-ayment for buying the drug. We can think of countries ith similar health needs and different social security systems. Some countries in the Euroean Union satisfy these characteristics. In this case, = =, and the condition for arallel trade to take lace is >. The total elfare can be reritten as: TW = TW (( 1 ) + ( 1+ ) ). Therefore, arallel imorts decrease the total elfare in this case, even hen the sum of the consumers surluses is ositive, hich occurs only hen:

24 >. + We have no a re-allocation of the drug consumtion from individuals in country ith a higher valuation:, +, toards individuals in country ith a loer valuation:,. This can be seen grahically in Figure 3, hile Proosition summarizes this case. Figure 3 Country : + Country : Proosition. Parallel imorts decrease the total elfare hen they take lace beteen countries differing only in their health insurance reimbursement olicies: > and = TW TW. 3

25 5. Conclusions With this aer, e articiate to the ongoing debate over the benefits and drabacks from alloing arallel trade among countries. We use a model that accounts for the differences beteen countries in terms of health system (reflected in the level of atients co-ayments), and in terms of drug needs (reflected in the atients valuation for the drug). Our main findings are the folloing. First, e confirm some results already discussed in the ongoing debate: Parallel trade makes the rices converge beteen countries, it makes the individuals of the imorting country better off, hile making the ones of the exorting country orse off, and they decrease the rofit of the monooly roducer. Moreover, e sho that the ublic exenses in both the imorting and the exorting countries are reduced ith arallel trade. Second, e sho that the effect of arallel imorts on the total elfare is ambiguous, even though the total outut remains unchanged. This certainly contrasts ith the classical result over the negative effect associated ith rice discrimination hen total outut is unchanged. This discreancy relies on the resence of differentiated co-ayments in our model. We identify to cases here the effect of alloing arallel trade on the total elfare can be stated unambiguously. On the one hand, e sho that arallel trade increases the total elfare hen it takes lace beteen to countries differing in their health needs only. The rationale behind this ositive effect relies on the re-allocation of the drug consumtion 4

26 from individuals ith relatively less drug needs in the exorting country, toards individuals ith relatively higher drug needs. On the other hand, e sho that arallel trade decreases the total elfare hen it takes lace beteen countries differing in their health system only. In that case, the drug consumtion is re-allocated from individuals ith relatively more drug needs to individuals ith relatively less drug needs. Our analysis is made maintaining the level of income equal beteen the countries. Therefore, our results are alicable to trade taking lace beteen countries of similar income levels. direct interretation of our results ould be the folloing: On the one hand, arallel trade ould increase the total elfare hen it takes lace beteen to develoing countries ith the same level of income and atients co-ayments, and different drug needs, to account for the higher needs for malaria or IDS treatment in some develoing countries than in other ones. On the other hand, arallel trade beteen industrialized countries, characterized by similar levels of (high) income and similar eidemiological conditions, and different drug reimbursement levels, ould decrease the total elfare. When e consider arallel trade beteen countries ith different income levels, such as the trade beteen develoing countries and develoed ones, e should carefully add the ell knon effects of arallel trade beteen a oor country and a rich country (re-allocation of the consumtion from the oor country toards the rich one) to the effects identified in the resent aer. 5

27 Given the results obtained in this aer, it ould be interesting to exlore the otimal decision of governments about hether to ermit arallel imorts or not, together ith an endogenous decision about the otimal health insurance reimbursement olicy. This additional ste is left for further research. In that sense, it is orth acknoledging the contribution of Richardson (00) ho demonstrates that, hen countries individually choose hether or not to rohibit arallel imorts, a global Nash equilibrium involves the ermitting of arallel imorting into all relevant foreign markets. References Danzon, P.M The economics of arallel trade. PharmacoEconomics 13 (3): Ganslandt, M. and K.E. Maskus Parallel imorts of harmaceutical roducts in the Euroean Union. Working aer No The research institute of industrial economics (Stockholm). Jelovac, I. 00. On the relationshi beteen the negotiated rices of harmaceuticals and the atients co-ayment. Working aer 008. Center for Economic Develoment. Hong Kong University of Science and Technology. Jelovac, I Monoolistic third-degree rice discrimination of subsidized goods. Mimeo. 6

28 Malueg, D.. and M. Schartz Parallel imorts, demand disersion, and international rice discrimination. Journal of International Economics 37: Maskus, K.E Parallel imorts. World Economy 3 (9): Maskus, K.E Parallel imorts in harmaceuticals: imlications for cometition and rices in develoing countries. Final reort to the World Intellectual Proerty Organization. Maskus, K.E. and Y. Chen. 00. Parallel imorts in a model of vertical distribution: Theory, evidence and olicy. Pacific Economic Revie 7 (): Pavcnik, N. 00. Do harmaceutical rices resond to otential atient out-of-ocket exenses?. RND Journal Economics 33 (3): Pecorino, P. 00. Should the US allo rescrition drug reimorts from Canada?. Journal of Health Economics 796: Richardson, M. 00. n elementary roosition concerning arallel imorts. Journal of International Economics 56: Tirole, J The theory of industrial organization. Cambridge: The MIT Press. 7